Kubernetes SIG CLI, 26 Jul 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes SIG CLI 20230726

Description

Kubernetes SIG CLI Bi-Weekly Meeting on July 26th, 2023.

Agenda and Notes: https://docs.google.com/document/d/1r0YElcXt6G5mOWxwZiXgGu_X6he3F--wKwg-9UBc29I/edit#bookmark=kix.p600f2h8fr6j

A

Hello and welcome to the July 26th edition of the six CLI bi-weekly meeting I'm Sean Sullivan I'm, your moderator I'll go directly into our announcements, um which is that the code is frozen, so you can't get any code in not until we release in August um and the the yeah they release date is scheduled to be Tuesday, August, 15th, so right around. Then this one code will unfreeze.

A

So now we, this is the time of the meeting where we get to introduce uh new members um and I know that there's at least one who uh who's put an item on the agenda. Is there anyone who would like to introduce themselves to the other 60 Liars.

A

Okay, why don't we move on so.

B

Wow should I say something or.

A

Sure sure now or later would be, and now is fine as well.

B

Oh okay, uh yeah I'm, just here I'm a new contributor here and I. uh The exploring open source I am a very extroverted person, make a.

A

B

From working with people, somehow I'm like 10 times as good, when I do and uh that's what built me to kubernetes and uh yeah, see how things go from there. I'm still looking for a first issue, if I can help someone or uh if there's something to do then uh yeah. If someone can, let me know yeah.

A

I think I think this should be a great opportunity to uh to uh to find out who else is in the Sig um and then also uh one of the better forums as well to to get a first item or first issue to work on will be the bug triage.

A

um That's generally a really good place to start and, of course, um Eddie uh zaneski. He runs that meeting, and so do you want to say anything. uh Eddie.

C

uh No welcome um I have a lot of catching up to do at the moment. The bug triage is definitely a good place to join us as well.

A

So the um I'm not sure if you already have, but if you, um if you join the Google Groups uh the email group, then you will be invited to all the six CLI meetings and I'll put that into the uh into the agenda shortly um and I'm. I'm sorry, but I um I didn't grab the pronunciation for your name. Is it Joshua.

B

You can just say yours is fine. Yes,.

A

B

It's a pleasure.

A

To meet you John, yes, thanks for coming to.

B

Join us likewise, I already got the things in my calendar, uh I think Eddie helped me or excellent.

A

A

Okay, great, we can move on to the sub project and cap updates.

D

Oh, can you hear me? Oh sorry, yeah.

A

Go ahead. Sorry.

D

Yeah I'm I'm, sorry, like I, don't know also just um used the occasion to to introduce myself um so uh yeah, so I'm, Andreas um I am working for red hat, like normally like more like partner, engineering and stuff, would be a little bit less like I'm, not a full-time developer um and I wanted to bring into a patches, but I realized that I did it all wrong and I should have like first joined the meetings um and then pick stuff and think and uh pick the items and stuff.

D

um So um I just wanted to bring like two things onto onto the agenda a bit later. um If it's possible um I, don't have I I requested added access, I, don't know who can give that to me, but otherwise I'll just do it in the open discussion without putting the items on the agenda.

A

So generally, it's um yeah I. Think that you might. You may have to join the the Google group in order to get the edit access.

D

Again, I think I put everything exactly the wrong way around so um Marie. Sorry about that. No.

A

No worries yeah, like I, said today's meeting is going to be pretty low-key. We, we didn't really have a whole lot on the agenda, and so we could be pretty flexible.

D

Great. Thank you.

A

Great is there anyone else that would like to introduce themselves.

A

If not, we can quickly note that at least one of the uh the Caps did not make the um the the transition from Speedy websockets um made some great progress. uh The cap has been accepted, but the code for the code freeze did not make it in, and so it looks like it's going to be pushed back to 129.

A

um and so we'll probably be working on both port forward and the remote command at the same time, or at least they will be released at the same time. That's the plan is that um the remote command code will be uh we're still working on it. As soon as the unfreeze happens, we could get that in and then start work on the port. Four.

A

We could actually on our note, we could start working on port forward um even sooner than that, if you like, I'm gonna I, think that at least one of the transition to websockets PRS is ready um and looks good uh according to Jordan and then we'll we'll get the the second one. The stream translator one hopefully, as you know, both of those as soon as the code freeze on unfreeze happens is.

E

A

Other kept that uh kept update or sub project update, we'd like to add here.

F

Interactivity, flag or delete comments was merged and it is in outside phase excellent.

A

Excellent thanks orda. Let me put that in.

C

It great job getting it over to finish. Linearta yeah.

F

C

One of the many things I started and never finished, I appreciate you.

F

Thanks I actually picked the work you have done and push it forward.

C

That makes me so happy. Thank you.

A

Thank you. So is there any uh sub-project update, maybe uh is there anyone from customize I would like to say anything or.

E

um I can give an update, uh so we launched our customized maintainer training cohort. Last week we had our first meeting um about 20 people showed up, um which is uh really looking good. um So hopefully some of those people will stick around and we'll get more maintainers.

A

A

Cool, so we could move on to the agenda items if, if there's no other sub-project or cap updates.

A

So I'm gonna skip the first item. Well, I. Think Yaz is already introduced himself um and go on to uh uh going to Justin's item.

G

uh Sure, thank you. So uh yes, uh this is based on the cap around I think we've made formula calling it prune V2, um uh but it's basically apply set functionality. uh We did get the outlet in Alpha into I, think 127.. So it's it's there. uh We discovered that there is at least one bug.

G

um This is a sort of odd one in that it is uh well in the cap. We actually said we were going to do something different than we what we implemented and it turns out that what we wrote in the kept. What we wrote in the kept was the right thing that we should have done, because what we implemented has a a regression where uh we cannot apply a crd in an instance of that crd in the same uh level apply yeah, because it can't predict the resource.

G

um I mean we could. We could make it work, but uh it would be pretty complicated logic. It would involve like reordering things or iterative type behavior, um and so uh this particular uh PR uh implements what we um specified in the cap and therefore would work around the issue. uh So, basically, switching from contains group resources to contains group kinds which I think is actually more user-friendly.

G

Anyway, um uh there was some feedback on the cap, which I think I was confused from like API Machinery folk, but I I think from a user-facing point of view, uh kinds is is better, um so most of it is actually just uh fixing the um the golden tests as it were the tests. uh One thing which I think is interesting is, although the feature is in Alpha, there is some limited accommodation so that we don't just like drop the previous value. We sort of try to migrate.

G

It and I actually wasn't sure whether we would you know what the right thing to do is for features of an alpha, whether we just say you know too bad so sad, uh it's alpha or whether we make some accommodation, um the obvious problem. Being at what stage should we drop this accommodation uh for users uh and I? It did not make feature, freeze or did not make feature, freeze argue, but it's a bug, so we could debate it or like as for exemption or whatever it's called.

G

If we, um if anyone cares enough, um but August 15th, isn't that far away so.

E

G

G

Like how much we try to not break people that are using feature flag to features.

A

So so I I think that it's it's kind of because it's Alpha there. There are no official um rules for accommodating people who are using the alpha um and I mean I. I wouldn't worry about it, um but.

G

It's it's pretty easy to work around, so it wasn't too bad, but yeah. The only problem is now we're like we have a harder decision, which is when do we remove the accommodation but I don't know make it beta or something like that.

A

C

Yeah, the prune thing is something that'll. Definitely a lot of people use it's something we hear about very often um we're gonna break people, no matter what we do and that's the reality of it.

A

And and but since it's Alpha I mean how many people do, you think really are or I mean it might be good to like try to get some feedback on that. Just to see how much, uh how much complaining or you know how much we're going to hear about it. But like you're right people do use the pruning but, like I, think the the Legacy pruning as opposed to this Alpha.

C

It's: how long has it been in Alpha? It's a real question right, a very long time right.

A

C

One of the Perma Alpha ones.

A

Now this is 127.

C

G

127. yeah: this is the new, the new Alpha, oh.

C

A

It's okay great! uh Should we move on to the the next item so varsha? Are you uh available.

H

Yeah um I just wanted to bring up this a super small PR which introduces the lender, as we discussed in the last community meeting to stop merging PRS with Anova FL flag. uh We wanted to get this in before 1.20 uh right before the code freeze for 1.27, sorry 28., but that didn't happen, I'm, not sure if it's so urgent. So it should be fine, but just mentioning here to get some ice so that we can watch this.

A

Okay, great so so it looks like there are some folks from the API Machinery, who are also uh reviewing.

E

A

Next, with Dr shamansky and we'll check, okay.

A

And Marley.

A

Okay, thanks for bringing that to our attention, um as you mentioned, uh where, if the code won't be, none of the code will be going in until August, uh but I think that is a a wise idea to to get this in front of people before then.

C

Yeah I think matcha can approve the hack directory. Otherwise we need someone else.

A

A

Then it looked like there was a release, note uh needed- or at least none put into the release note, um although we we may want to release note with this linter change, foreign.

H

A

Usually it's in the template for the pr they'll, be release node and then triple backslash backslashes on two different lines um and um and then you'll. If, if there are, is no release, note you just put in all capital letters none in between those triple quotes, um and so, if you, if you just open another PR you'll, see it in the template for the release note and then you can copy that. Don't fully open a PR, but you know pretend to open one you'll see the the release.

A

Note template I I could show you I'll contact you um afterwards.

H

I can do that.

A

So so, let's move on to the the next PR that we have here which says Andreas so Andreas. Would you like to to drive.

D

Yeah, so um it's technically, uh it's a bug fix right. So again, it's probably the wrong meeting for this, but like um it's, uh it's like a so again if it's all wrong and stuff and now backwards.

D

um It's an alternative proposal to like a fix that Brian had like tried um about half a year ago and then in the meantime, another PR emerged and um Brian's approach like didn't work anymore.

D

um So in order to fix that bug like to uh yeah, it's too, it's basically parse complex, Json, Expressions right, so the weight four and the weight four like um okay, currently like it, breaks on single, um equal signs and then breaks the string apart and then evaluates Expressions, um and that doesn't work because, when you're like sub selecting right because the Json path Expressions um have like double equal signs to to mention a condition.

D

um So what I'm trying to do with the pr is just um to basically ignore double equal signs and only on only break on single on single equal signs, and so that way like it would be possible to sub-select um like on the condition right when you like matching conditions and um basically going in like on the status fields. For example, um and sub selecting on them.

A

Okay, yeah I, think I've heard several um I've heard this issue several times and it sounds like uh we might have stomped on each other. A couple of the PRS here is that what you're saying yeah.

D

Exactly so it's um it's! My sorry are you uh just? Are you Brian or no.

A

No right Brian's here, though, I think oh okay, yeah.

C

I'm here yeah I stumped.

E

F

Know I did see.

C

Your comment on my other, so I tried to rebase my PR, but the changes that I made I made the changes was made. It really difficult for me to rebase what I had already done so um so I'm happy to take a look at your PR and I um I. Have it on my list of things to do I'll, try to take a look at it today and leave some comments if there's anything um but yeah, I, I I think it's fine to go ahead with with your PR.

C

If, assuming that it passes the test cases that I that I mentioned. That's my only concern because I know there were some kind of tricky scenarios that that I need to be able to handle. So I'll take a look at that and leave some feedback, and if anybody else does feel free to also look at it as well sounds.

D

Good and I know that I know that I've stomped on your feet without once I'm totally fine, also to not get it in right like it's like no I should have joined the meetings. First figured out that if there was a bug fix for this already or not and then taking on the work and I just did it the other way around, because I don't know like it's. No.

C

You're fine I mean right, the pr was open for a long time and started getting stale, so it definitely um needed to be looked at again anyway. So I'm fine with it I'll. Look at your PR, okay. Thank you so much yeah. Thank you for contributing and joining.

D

C

We are stoked no, no toe stepped on for sure. uh Marley I also tagged you on this as our other Jason path, expert, I'm, sorry, no problem thanks, yeah I've, looked at it a couple of times. I just haven't gotten a chance to sit down and really dig into it.

A

Cool and so the the next one Andres is uh about Coupe cuddle weight, yeah.

D

So it's it's a similar, so it's both both are related to like the same work. um So this comes like from an issue where, like somebody, couldn't wait on or wanted to wait on multiple conditions, um either either horde or like or like or logical or logical, end, and then um I think Marley suggested that one might add like a dash, multiple dash, dash, fours and then perhaps the dash dash for all or something like that.

D

um I mean I, I'm, not really sure. If I know this would need a cap, and this would need all of the other bells and whistles like this, come with it again. I did it the wrong way around, um but but the other thought that I have is um so so either right either. The the parsing for the for the Json path.

D

Expression would need to get better and and like one would add, like perhaps like a an or character or something to to make sure that inside the same string, one can wait for several different conditions right or or one would do it with like these several different command line, Flags.

D

um So I don't really know. What's like the. So this is basically like a POC for how this could look like or that and it works, um but uh but I'm not sure if it's like the best way to wait for multiple conditions, if it's even worthwhile doing it um and with all the problems with the problems coding code right with the Json public Expressions, um it wouldn't be better to change them but I.

D

Suppose if you change them, then you get like a or if you change the parsing right and if you remove press the lazy, parsing, parsing or stuff like that, then probably break the user-facing interface. So that's also difficult.

D

um So it's I, don't know it's it's just so. Basically my question is: does it make sense to add like support for waiting for multiple conditions, um either a logical and and logical, logical or or is this Overkill?

D

um And so this was basically just my attempt to fix that that feature request.

A

Yeah I think that this is quite a bit to uh bite off um and will require uh coordinating with the folks that are already working on the weight um and it does sound like um the it would, especially if we're changing the flags that it would probably require a cap.

A

um So so this just so you know this. This would be a significant effort um if, in fact, that's what we, the the path that we think is, is useful and and you're willing to to pursue that path.

D

I would yeah so I'm gonna have no problem either abandoning this or working on this photo right or like it. So that's um for me, for me: that's that's either or um again it was kind of tinkering around, but it was just like lying around and was just like. Oh I can just try this right, but again, like neither of these PRS I have to absolutely get in I.

D

Think just the experience also for for having this feedback now and for talking about this and figuring out how all of this works is also it's also worthwhile. So I might also, perhaps just in the future, see if I pick up something somehow some something else or like, like the like an earlier participant set right like that, just just to get started with it and understanding how these things work.

A

So yeah I think that it'd be important for anybody who's already working in this area, Marley and Brian. If they have any ideas on whether or not it's useful to try to get multiple uh conditions for weight, foreign.

A

Does anybody have any opinions on the usefulness of that.

C

I mean I definitely think it's useful.

C

um The way that our Json past stuff works today might make it incredibly difficult to make work, but you know, since we don't support so much of it.

C

um I think that the easiest way with where we're at currently would be to just have multiple uh four Flags get broken out into multiple Watcher loops, um but I, don't even know I, don't know. I haven't I haven't looked into it much beyond that.

D

So so that's kind of what so I.

D

That's what you suggested also on the on the issue right and so like that was kind of what I tried to implement might be incredibly ugly, to be honest but like so what I tried is to keep the existing code and basically just put it into um like into like uh into like several go functions right and then, and basically then just check depending on, if it's like a for all or not like, if the first one returns positively or one of them returns positively, then uh then return true right or depending on the flag right, um so yeah I, don't I, don't know if you want.

D

If, if you have cycles of your I, don't know how this works right, but if you want to or have cycles and just want to have have to look at the at the at the pr and like consider this just as a POC and and see if this is something that could be feasible and um um and then perhaps I don't know what what would take it would take a cabin stuff right and everything everything else.

C

Okay, yeah I think like Sean. If we um we made it so you just like it, doesn't really change the baseball Behavior we just like, if you add more fours and you added like an extra flag. That said, like um I, don't know something that indicated an and or an or those would all be additions, and thus wouldn't really need a cap or any any of the like longer Cycles to to not break users. I think right.

A

Okay, so it would just be an assumption that more Flags would be ands.

C

A

C

Could do it if we're just like having more Flags like there would be a default state of probably being and and then an extra flag could be passed to define whether it's an and or an or of course you can't really do much more complicated than that, but at the same time like at that point, I think it just becomes like plug-in territory.

C

Well, I think you could I mean really. You could simulate and just by doing multiple, multiple cuddle weights right like if, if you, if you support or then you could just do you know a wait for.

F

E

True so options you don't even.

C

Need to support and really yeah.

C

E

Basically, it's just a.

C

Good point because then, then we just fall back on basic basic shell functionality, so it's just the aura that it needs to be oh God, yeah I know I meant to try and import an existing third-party library, but it just didn't get very far with it. Custom.

G

uh I you've teamed me up perfectly uh on that. No, my understanding is that Json path itself isn't like terribly well defined and there are other things like Jane J mess path, but also there's cell, uh like we don't have support for cell today in Cooper, if I understand correctly, but I know that the API Machinery is doing it, pushing it pretty hard I, don't know if we have a view of whether sellers sufficiently user friendly to go into Coupe cuddle, whether it's you know crd author's, only type thing uh but yeah.

G

Maybe that is a I mean if we're gonna have to do a cap any well it's available yet, but if we're gonna have to do a cap anyway, maybe like adding cell support, is something to consider.

A

Yeah there's they they had to compile in that that cell runtime right into the API server I mean that so that would be something to consider. It's like an entire run time into the cuddle.

G

Yeah but you're going to build it incrementally otherwise, right, like so yeah.

A

Yeah, it is yeah good point not.

C

Having a full Jason path has bitten us so many times and caused so much friction and issues for users so yeah the cell is confusing as hell. Every time I have to go, do gcp, I am principal stuff, so yeah Philip, you have your hand ready. Sorry.

I

Yeah I just want to add that there is a longer standing issue regarding this and I heard from some people to ask for this feature as well. So um maybe we can reference this issue also in in the pr that.

C

A

E

A

C

A

Should we add that to the agenda, if possible, a link to that uh issue? Philip yeah.

I

Sure I can add it there as well great.

A

Thanks yeah, it feels like uh there's been quite a few of these Json issues hanging around for the last nine months or so.

C

Api Machinery did go through with the cell change, though with I guess that went through a cap.

A

Yeah it so they have it for for validation for uh validating web books or validation within the API server itself and they're. Also they haven't done mutating uh web hooks or mutating functionality through it yet, but that apparently, is on the horizon. That's my understanding. Does anybody else have a different understanding of the current state of cell.

A

Yeah I know that there's multiple people uh that I'm working with that I keep hearing uh cell work about with.

C

And someone pays what y'all are talking about under the chat with query is hard. Thank.

H

C

I think this is the first one I found.

G

Yes, I'd also mention uh like Eddie to your point. I think uh I've also found self use because of the behavior around optionals I believe that they've done some special handling so that it is more tolerant of missing fields or something like that.

G

So it's a little bit less fussy, then.

E

G

E

I am providers.

C

But yeah I I think I'm, definitely a big plus one, and we should road map to do something about Jason path, whether that's actually get a full two spec and the problem is being that the Json path. Spec is super light and what everyone knows as Jason path or like utility functions that certain libraries have sprinkled in.

C

So we support I'm, pretty sure we support the actual spec of jsonpass. We just don't have the utility functions that people are.

F

C

We don't support the.

E

Channel we don't.

C

We don't support the full spec we support like 75 of it. Okay, thank you for fact. Checking me.

A

Okay, well, that turned out to be a much more fulsome meeting than I I thought and that, according to the initial agenda, but uh thanks for bringing all of those issues into the agenda and for showing up is there anything that we'd like to talk about before we sign off.

A

I

F

I

B

Sure I was wondering if you may know, of another open source project that has like a very social Community like kubernetes or.

A

I know that there's quite a few uh open source communities- I I'm, not as uh involved in in many of those, but maybe some other people have ideas on uh social open source projects.

A

This is this is the one that I I contribute to the uh the most.

A

So this there there is a a uh the cncf, the cloud native Computing Foundation, which is the umbrella for multiple open source projects um and, of course, Linux is, is a very uh well-known open source project um and there's tons and tons of them.

A

um Does anybody have any ideas- and you say, you're looking for the more social ones.

B

Yeah, where you have like a slack Channel people are active on it, that's not common yeah. This is what I saw.

C

Yeah we, the kubernetes project, has tons of different sigs too. um So definitely don't get scared away. Right now, like you could definitely hang around with us. Work will definitely pop up to be done, uh but there's other cigs and other subgroups and sub-projects too. So there's there's definitely a home for you within kubernetes, and we can. We can help facilitate that. We need all the contributors we can get.

C

uh Sorry I have one last thing, so we actually have a deadline coming up for the uh one minute update video that we get for the kubecon main stage. So we talked about this I think last meeting.

C

um Okay, the I forgot to put this on the agenda. We need to record August. 10Th is the deadline, so we need to record our one minute update for six CLI. This is a rad opportunity for us to get in front of the entire Cloud native con kubecon audience during the keynote and talk about the cool work we did and the stuff we're doing and what we need contributors for so are there. I could pull some of the stuff that made it into 128, which is cool.

C

We can update any other work that we want to make sure we shout out any other in progress efforts that we want to let the the rest of the world know about, and if you don't think of them now you can drop them in the agenda in the dock or slack them to us. But we will probably record that probably the day before it's due because that's how we usually operate. But oh.

A

Cool thanks for bringing that up. Eddie I've forgotten about that foreign.

A

A

I'll give you guys 20 more minutes back, okay, thanks for joining us, and we will see you again in two weeks, foreign.